> Home >

HadCRUH: Version Differences

The most recent version of HadCRUH was created in July 2008 and is the currently available dataset. The only way it differs from the original version of HadCRUH is through the process of blending the land and marine gridded data. Only coastal stations (where both land and marine data are present) are affected. In the original version of the dataset due to programming errors, coastal grid-boxes contain only marine data. This has been corrected and now a weighted average of the land and marine data present in each grid-box is used (as documented in the published literature). Components are weighted by their proportional area of each grid-box but with a caveat of a 25% minimum to ensure sufficient representation of both components where traditionally (for temperature at least - Jones et al. 2001) small island land data was thought more reliable than marine data.

Differences

The original and new datasets have been compared both at grid-box and zonally averaged scales and both in terms of timeseries and trends. The results are summarised below. The mean difference in the specific humidity grid-box timeseries is -0.001 g/kg and the standard deviation of the differences is 0.001 g/kg. For relative humidity the mean difference is -0.003 % and the standard deviation of the differences is 1.066 %.

All differences are found to be very small and do not affect the broad conclusions drawn from analysis of this product. For comparison, the original version of HadCRUH can be found here: HadCRUH Original Version Download

DIFFERENCES IN SPECIFIC HUMDITY

Difference Between Original and New Decadal Trends of Specific Humidity in g/kg for the Period 1973 to 2003

Original minus new HadCRUH Decadal trends of surface specific humidity are shown in g/kg over the period 1973 to 2003. Maximum differences are +0.19 g/kg and -0.26 g/kg. Mean difference is -0.0001 g/kg and standard deviation of the differences is 0.001 g/kg.

Difference between Original and New Zonally Averaged Monthly Mean Anomaly Surface Specific Humidity and Decadal Trends in g/kg

New zonally averaged monthly mean anomaly (relative to a 1974 to 2003 climatology) surface specific humidity timeseries for the Globe (70 °N to 70 °S), Northern Hemisphere (70 °N to 20 °N), Tropics (20 °N to 20 °S) and Southern Hemisphere (20 °S to 70 °S) are shown in red. Original minus new timseries are shown in blue. Both original and new decadal trends are also shown where ** and * denote significance at 1 % and 5 % respectively. The trend was estimated using the Restricted Maximum Likelihood method. Mean differences and standard deviation of the differences for the Globe, Northern Hemisphere, Tropics and Southern Hemisphere are: -0.001 and 0.021, -0.00009 and 0.028, -0.002 and 0.033 and -0.002 and 0.017 in g/kg.

DIFFERENCES IN RELATIVE HUMDITY

Difference Between Original and New Decadal Trends of Relative Humidity in % for the Period 1973 to 2003

Original minus new HadCRUH Decadal trends of surface relative humidity are shown in % over the period 1973 to 2003. Maximum differences are +0.93 % and -0.72 %. Mean difference is -0.00002 % and standard deviation of the differences is 0.005 %.

Difference between Original and New Zonally Averaged Monthly Mean Anomaly Surface Relative Humidity and Decadal Trends in %

New zonally averaged monthly mean anomaly (relative to a 1974 to 2003 climatology) surface relative humidity timeseries for the Globe (70 °N to 70 °S), Northern Hemisphere (70 °N to 20 °N), Tropics (20 °N to 20 °S) and Southern Hemisphere (20 °S to 70 °S) are shown in red. Original minus new timseries are shown in blue. Both original and new decadal trends are also shown where ** and * denote significance at 1 % and 5 % respectively. The trend was estimated using the Restricted Maximum Likelihood method. Mean differences and standard deviation of the differences for the Globe, Northern Hemisphere, Tropics and Southern Hemisphere are: -0.004 and 0.049, -0.0003 and 0.096, -0.003 and 0.056 and -0.01 and 0.066 in %.

Dataset produced in collaboration with: